In [1]:
import numpy as np
import pandas as pd
import skimage.color as color
import os
total_files=[]
fp="/wheat/train/"
for dirname, _, filenames in os.walk(fp):
total_files=filenames
In [2]:
root_path = "wheat/"
train_folder = os.path.join(root_path, "train/")
test_folder = os.path.join(root_path, "test/")
train_csv_path = os.path.join(root_path, "train.csv")
sample_submission = os.path.join(root_path, "sample_submission.csv")
In [3]:
train_df=pd.read_csv(train_csv_path)
In [16]:
from skimage.filters import laplace
import numpy as np
import skimage.filters as filter
import matplotlib.pyplot as plt
from skimage.filters import laplace
from skimage import io
fig,ax=plt.subplots(1,2,figsize=(24,24))
for en in range(2):
image = io.imread(train_folder +str(train_df.loc[0,"image_id"])+".jpg")
if en==0:
ax[en].imshow(image)
continue
im = color.rgb2gray((image))
im1 = np.clip(laplace(im) + im, 0,1 )
ax[en].imshow(im1)
In [5]:
import skimage.segmentation as seg
import skimage.filters as filters
import skimage.draw as draw
import skimage.color as color
import skimage.transform as tfm
def image_show(image, nrows=1, ncols=1,cmap="gray"):
fig, ax = plt.subplots(nrows=nrows, ncols=ncols, figsize=(8, 8))
ax.imshow(image,cmap="gray")
ax.axis('off')
return fig, ax
fig,ax=plt.subplots(1,1)
ax.hist(image.ravel(),bins=64,range=[0,300])
ax.set_xlim(0,300)
Out[5]:
In [6]:
fig,ax=plt.subplots(1,5,figsize=(12,12))
for en,i in enumerate([0,50,60,80,100]):
image = io.imread(train_folder+str(train_df.loc[0,"image_id"])+".jpg")
if en==0:
ax[en].imshow(tfm.resize(image,(512,512)))
continue
mask=image<i
image[mask]=0
gimage=color.rgb2gray(image)
im = color.rgb2gray((image))
im1 = np.clip(laplace(im) + im, 0,1 )
ax[en].imshow(gimage,cmap="gray")
#image_show(fimage)
In [7]:
t_filter=[filters.threshold_li,
filters.threshold_triangle,
filters.threshold_yen,
filters.threshold_isodata,
filters.threshold_sauvola
]
fig,ax=plt.subplots(1,5,figsize=(32,32))
fig.tight_layout(pad=3.0)
for en,i in enumerate(t_filter):
tit=str(i)
image = io.imread(train_folder+str(train_df.loc[0,"image_id"])+".jpg")
i=i(image)
mask=image<i
image[mask]=0
gimage=(color.rgb2gray(image))
ax[en].imshow(gimage,cmap="gray")
ax[en].set_title(tit,fontsize=20)
In [8]:
t_filter=[
filters.threshold_mean,
filters.threshold_minimum,
filters.threshold_niblack,
filters.threshold_yen
]
fig,ax=plt.subplots(1,4,figsize=(22,22))
fig.tight_layout(pad=3.0)
for en,i in enumerate(t_filter):
tit=str(i)
image = io.imread(train_folder+str(train_df.loc[0,"image_id"])+".jpg")
i=i(image)
mask=image<i
image[mask]=0
gimage=(color.rgb2gray(image))
ax[en].imshow(gimage,cmap="gray")
ax[en].set_title(tit)
In [9]:
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
from skimage import data
from skimage.filters import threshold_yen
from skimage.segmentation import clear_border
from skimage.measure import label, regionprops
from skimage.morphology import closing, square,rectangle,convex_hull_object
from skimage.color import label2rgb
def make_rect(name):
n_image = io.imread(test_folder+str(name))
image = color.rgb2gray(n_image)
thresh = filters.threshold_yen(image)
bw=closing(image < thresh, square(4))
label_image = label(bw,connectivity=1,background=255)
image_label_overlay = label2rgb(label_image, image=image, bg_label=0)
fig, ax = plt.subplots(figsize=(10, 6))
regionprops(label_image)
ax.imshow(n_image)
listrect=[]
for region in regionprops(label_image):
# take regions with large enough areas
if region.area >= 500:
# draw rectangle around segmented coins
minr, minc, maxr, maxc = region.bbox
l=(maxc-minc)*(maxr-minr)*0.0264583333
listrect.append([int((l//10)*0.0264583333)])
rect = mpatches.Rectangle((minc, minr), maxc - minc, maxr - minr,
fill=False, edgecolor='red', linewidth=2)
ax.add_patch(rect)
return listrect
In [10]:
d=dict()
fp=test_folder
for dirname, _, filenames in os.walk(fp):
total_files=filenames
data_list=list(set(train_df["image_id"].values))
for k in range(len(data_list)):
data_list[k]=data_list[k]+".jpg"
notthere=[]
for i in filenames:
if i not in data_list:
notthere.append(i)
print(len(notthere))
In [17]:
temp_df=pd.DataFrame({"image_id":[],"bbox":[]})
for i in notthere:
rects=make_rect(i)
rects=rects
rects=" ".join(str(item) for innerlist in rects for item in innerlist)
d={"image_id":[],"bbox":[]}
d["bbox"].append(rects)
d["image_id"].append(i[:-4])
# print(pd.DataFrame(d))
temp_df=temp_df.append(pd.DataFrame(d),ignore_index=True)
In [12]:
temp_df["bbox"]
Out[12]: